Pith. sign in

REVIEW 3 major objections 4 minor 2 cited by

MOL: Joint Estimation of Micro-Expression, Optical Flow, and Landmark via Transformer-Graph-Style Convolution

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read Jointly training micro-expression recognition with optical flow and landmark detection lets one network read subtle facial actions from raw video, beating prior methods without key frames or hand-crafted features.

desk verdict A useful architecture with solid ablations, but the SOTA claim is overbroad and the Aff-Wild2 pretraining confound means the reported gains over prior work aren't cleanly attributed. read the letter →

arxiv 2506.14511 v1 pith:QLPDKID7 submitted 2025-06-17 cs.CV

classification cs.CV
keywords micro-expressionrecognitionmulti-tasklearningopticalflowestimationfaciallandmarkdetectiontransformer-styleconvolutiongraphF5Cblockrawvideoend-to-end
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to show that facial micro-expression recognition can be solved end-to-end from raw frame sequences, without needing key frames or pre-extracted features, by training the recognition task together with two auxiliary tasks: optical flow estimation and facial landmark detection. The claim is that these auxiliary tasks teach the shared feature extractor about the subtle muscle movements that define micro-expressions, and that this shared learning compensates for the small, low-diversity training datasets in the field. On the three standard benchmarks (CASME II, SAMM, SMIC) the resulting framework, MOL, reports the best or second-best accuracy in most single-dataset, composite, and cross-dataset evaluations, with the best three-class accuracy on CASME II (91.26) and SMIC (80.71). A sympathetic reader would care because the method removes two practical bottlenecks of prior work, and the ablation study attributes the gains to the joint learning and the new feature block rather than to extra supervision.

What carries the argument

The carrying mechanism is the F5C block, a plug-and-play feature extractor made of two operations. Fully-connected convolution (FCC) treats each row and each column of the feature map as a patch, adds positional embeddings, and applies circular convolution along both directions, so each output position sees the whole row or column, giving local detail with a global receptive field in the style of a transformer. Channel correspondence convolution (CCC) builds a k-nearest-neighbor graph whose vertices are feature-map channels, computes edge features from the global channel vector and the difference to each neighbor, and aggregates them with max pooling to model correlations among feature patterns. The F5C features are shared by the MER classifier, a FlowNet-style optical-flow decoder, and a landmark regressor; consecutive-frame feature concatenations are fed to a 3D CNN for the final classification.

What would settle it

Retrain MOL on the same benchmarks but replace the pseudo-label generators with human-annotated landmarks and optical flow, or with a different landmark and flow model; if the accuracy gap over the no-auxiliary-task variant shrinks or vanishes, the reported gains depend on the specific pseudo-label tools rather than on multi-task learning per se.

Watch

Extended reading notes

Core claim

MOL's central claim is that multi-task learning with automatically generated motion and shape supervision is enough to make a deep network attend to micro-expression cues that are barely visible in raw pixels. The framework processes a clip of raw frames, extracts a shared local-global feature for each frame with the F5C block, and from those features simultaneously predicts the micro-expression category, the optical flow between consecutive frames, and the facial landmarks of each frame. On CASME II, SAMM, and SMIC the authors report that MOL outperforms state-of-the-art methods in most configurations, including the best three-class accuracy on CASME II (91.26) and SMIC (80.71), and that the two auxiliary tasks and the F5C block each contribute substantially in ablations. The paper also reports that the auxiliary tasks improve each other: adding MER and landmark detection lowers optical-flow error, and adding MER and optical flow lowers landmark error.

Load-bearing premise

The load-bearing premise is that the automatically generated pseudo-labels, Dlib's 68 landmarks and TV-L1 optical flow, are accurate enough to teach the shared features about real micro-expression motion even though these tools were not designed for micro-expression videos.

Editorial extensions

If this is right

  • MER can be deployed on raw video clips directly, so pipelines no longer need onset/apex frame detectors or hand-crafted optical-flow inputs.
  • Because optical flow and landmark estimation improve in the joint framework, the same network can serve as a single model for several face-analysis tasks on small datasets.
  • The claimed results imply that pseudo-labeled auxiliary tasks can substitute for large-scale human annotation in micro-expression research, at least when the labels come from reliable off-the-shelf tools.
  • Cross-dataset gains (training on CASME II or SAMM, testing on SMIC) suggest the auxiliary tasks improve generalization to unseen subjects and recording conditions.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the pseudo-label dependence is real, then replacing the automatic landmark and flow generators with human-refined or learned supervision could push accuracy higher, and worse pseudo-labels could erase the gain; this is testable by swapping label generators.
  • The pair-feature concatenation effectively treats every consecutive frame pair as a mini-clip; an untested corollary is that the same architecture may extend to micro-expression spotting, where onset and offset are unknown, by classifying all pairs and thresholding confidence.
  • The FCC's circular convolution with positional embeddings is a lightweight alternative to full self-attention; one could adapt the block to other fine-grained video tasks such as action unit detection or subtle gaze change, where local detail and global context both matter.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. This paper proposes MOL, an end-to-end framework for micro-expression recognition (MER) that jointly learns optical flow estimation and facial landmark detection as auxiliary tasks. The architecture uses a novel F5C block combining fully-connected convolution (FCC) and channel correspondence convolution (CCC), operating on raw frame sequences without key-frame annotations. Experiments on CASME II, SAMM, and SMIC show improvements over several prior MER methods in most settings, and ablations demonstrate the contribution of each proposed component and auxiliary task. The paper also reports competitive performance for optical flow and landmark prediction.

Significance. The work is a well-executed empirical study of a novel multi-task architecture for MER. Its strengths include a clean ablation design (Tables 8–12) that consistently shows the F5C block, FCC, CCC, and auxiliary tasks each contribute to accuracy; a clear end-to-end pipeline that does not require hand-crafted features or key frames; and public code release. The main claims are falsifiable through the reported tables. However, the unablated Aff-Wild2 pretraining and the use of pseudo-labels as ground truth for auxiliary-task evaluation leave some empirical questions unresolved, so the paper's significance depends on the authors' ability to address these points in revision.

major comments (3)
  1. [Abstract, Sec. 4.2.1, Tables 4 and 7] The abstract and conclusion claim that MOL 'outperforms the state-of-the-art MER methods on CASME II, SAMM, and SMIC benchmarks,' but this wording is not supported by the reported numbers. In Table 4, MERSiamC3D achieves 81.89 Acc versus MOL's 79.23 on the five-class CASME II setting, and in Table 7 AU-GACN achieves 45.10 Acc versus MOL's 44.58 on SAMM→SMIC. Sec. 4.2.1 also concedes that GACNN, MERSiamC3D, and I2Transformer outperform MOL in a few cases. The claim should be qualified to state that MOL achieves state-of-the-art or competitive results on most settings, or the comparison should be restricted to the settings where MOL is strictly best.
  2. [Sec. 4.1.3] Every reported experiment pre-trains MOL on Aff-Wild2, but the paper never ablates the pretraining stage. The internal ablations in Tables 8–11 show that the F5C block and auxiliary tasks improve accuracy conditional on this pretraining, but they do not quantify how much of the margin over prior methods comes from Aff-Wild2 transfer (including its Dlib/TV-L1 pseudo-supervision). Please add an experiment that trains MOL from scratch or with a matched pretraining protocol to isolate the contribution of the proposed architecture and joint losses; otherwise the headline SOTA claim is underdetermined by the current evidence.
  3. [Sec. 4.1.1, Tables 14–15] The optical flow and landmark evaluations use TV-L1 and Dlib outputs as ground truth, and MOL is trained to predict exactly these pseudo-labels. The comparisons against RAFT, UnsupFlownet, TCDCN, and HRNetV2 are therefore not apples-to-apples unless those baselines are fine-tuned on the same data and evaluated against the same pseudo-ground-truth. The numbers in Tables 14–15 should be framed as 'agreement with TV-L1/Dlib' rather than as general optical-flow/landmark accuracy, or the evaluation should be repeated on standard benchmarks with human-annotated ground truth.
minor comments (4)
  1. [Table 4 header] The dataset name is misspelled as 'CAMSE II'; it should be 'CASME II'.
  2. [Sec. 4.3.2] The proposed block is referred to as 'FC5' instead of 'F5C' in the phrase 'one FC5 block is already sufficient'; please correct this typo.
  3. [Sec. 4.2.2] The statement that MOL is 'the only one DL based method with raw frame images as input' in the composite evaluation is misleading, since MicroNet, FRL-DGT, and SelfME also take raw images as input; the intended distinction appears to be that those methods require key frames. Please clarify the wording.
  4. [Sec. 4.4] The paper does not report the computational cost or parameter count of MOL; a brief efficiency comparison with prior methods would be useful, especially since the proposed FCC and CCC add global operations over the full feature map.

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity; results are empirical supervised evaluations, with minor non-load-bearing self-citations and a disclosed pseudo-label/pretraining confound.

full rationale

The paper's derivation chain is an empirical end-to-end learning pipeline, not a formal derivation. The central MER results (Tables 4-7) are obtained by leave-one-subject-out cross-validation on held-out subjects, so accuracies are genuine test-set predictions rather than fitted constants. The auxiliary-task losses L_f and L_m use TV-L1 and Dlib outputs as pseudo ground truths (Sec. 4.1.1); this is a disclosed weak-label choice, and the auxiliary predictions are evaluated against the same pseudo-labels (Secs. 4.4.1-4.4.2), which is a benchmark-validity caveat but not a circular reduction of the MER claim: the main classifier is trained on human ME labels. The hyperparameters (k=4, lambda_f=0.1, lambda_m=68) are tuned on CASME II validation and fixed elsewhere, which is standard practice. Self-citations [25], [37], [38] are used for comparison and for inter-ocular normalization provenance; none is load-bearing, and no uniqueness theorem or ansatz is imported from the authors' prior work. Two concerns flagged in the manuscript do not constitute circularity: (i) Aff-Wild2 pretraining (Sec. 4.1.3) is never ablated, so the marginal contribution of F5C and the auxiliary tasks over pretraining is underdetermined; this is a missing control, not a self-referential derivation. (ii) On 5-class CASME II, MERSiamC3D reports 81.89 Acc vs MOL 79.23 (Table 4), weakening the 'outperforms state-of-the-art' wording; this is a correctness or claim-calibration issue, not circularity. Overall, no equation reduces to its own input by construction, and the reported results are externally benchmarked against held-out subjects.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central performance claim depends on standard supervised learning assumptions plus two domain-specific choices: the reliability of automatically generated pseudo-labels and the transferability of Aff-Wild2 pre-training. Both are plausible but not independently validated in the paper.

free parameters (4)
  • lambda_f = 0.1
    Weight for optical flow loss in Eq. 9; set manually, no sensitivity analysis reported.
  • lambda_m = 68
    Weight for landmark detection loss in Eq. 9; set manually.
  • k = 4
    Number of nearest neighbors in CCC graph; selected by validation on CASME II five-class and fixed for all other datasets.
  • t = 8
    Number of frames per input clip; Table 13 shows performance varies with t, chosen without clear criterion.
assumptions (4)
  • domain assumption ME class labels in CASME II, SAMM, and SMIC are correct and consistent
    Labels are taken from the datasets and used for cross-entropy loss; see Sec 4.1.1.
  • domain assumption Dlib and TV-L1 provide reliable pseudo ground truths for landmarks and optical flow
    Both are used as supervision for auxiliary tasks; see Sec 4.1.1 and Eqs. 7-8.
  • domain assumption Pre-training on Aff-Wild2 transfers to micro-expression recognition
    MOL is pre-trained on Aff-Wild2 then fine-tuned; no ablation of this step is provided (Sec 4.1.3).
  • standard math LOSO cross-validation provides unbiased performance estimates
    Adopted from prior MER works; see Sec 4.1.2.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MOL: Joint Estimation of Micro-Expression, Optical Flow, and Landmark via Transformer-Graph-Style Convolution." pith.science (2026). https://pith.science/paper/QLPDKID7

@misc{pith2026250614511,
  author       = {Pith},
  title        = {Pith review of: MOL: Joint Estimation of Micro-Expression, Optical Flow, and Landmark via Transformer-Graph-Style Convolution},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QLPDKID7}},
  note         = {Machine review of arXiv:2506.14511}
}
read the original abstract

Facial micro-expression recognition (MER) is a challenging problem, due to transient and subtle micro-expression (ME) actions. Most existing methods depend on hand-crafted features, key frames like onset, apex, and offset frames, or deep networks limited by small-scale and low-diversity datasets. In this paper, we propose an end-to-end micro-action-aware deep learning framework with advantages from transformer, graph convolution, and vanilla convolution. In particular, we propose a novel F5C block composed of fully-connected convolution and channel correspondence convolution to directly extract local-global features from a sequence of raw frames, without the prior knowledge of key frames. The transformer-style fully-connected convolution is proposed to extract local features while maintaining global receptive fields, and the graph-style channel correspondence convolution is introduced to model the correlations among feature patterns. Moreover, MER, optical flow estimation, and facial landmark detection are jointly trained by sharing the local-global features. The two latter tasks contribute to capturing facial subtle action information for MER, which can alleviate the impact of insufficient training data. Extensive experiments demonstrate that our framework (i) outperforms the state-of-the-art MER methods on CASME II, SAMM, and SMIC benchmarks, (ii) works well for optical flow estimation and facial landmark detection, and (iii) can capture facial subtle muscle actions in local regions associated with MEs. The code is available at https://github.com/CYF-cuber/MOL.

Figures

Figures reproduced from arXiv: 2506.14511 by the authors.

Figure 1
Figure 1. Illustration of optical flow and facial landmark differences between [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The architecture of our MOL framework. Given a sequence of [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The structure of our proposed transformer-style fully-connected [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: The structure of the optical flow estimation module, which con [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Visualization of optical flow estimation results for three example frame pairs [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: Facial landmark detection results for example images from [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Boosting Micro-Expression Analysis via Prior-Guided Video-Level Regression

    cs.CV 2025-08 conditional novelty 5.0 of 10

    A prior-guided video-level regression with adaptive interval selection and full parameter sharing sets new state-of-the-art results on micro-expression spotting and recognition benchmarks.

  2. Straightforward Bayesian A/B testing with Dirichlet posteriors

    stat.ME 2025-08 unverdicted novelty 3.0 of 10

    The submission is internally inconsistent: the abstract promises a Bayesian A/B testing method, but the full text is a different computer vision paper, leaving the claimed result unevaluable.

Reference graph

Works this paper leans on

69 extracted references · 66 canonical work pages · cited by 2 Pith papers

  1. [1]

    Video-based facial micro-expression analysis: A survey of datasets, features and algorithms,

    X. Ben, Y. Ren, J. Zhang, S.-J. Wang, K. Kpalma, W. Meng, and Y.-J. Liu, “Video-based facial micro-expression analysis: A survey of datasets, features and algorithms,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 9, pp. 5826–5846, 2022

  2. [2]

    Ekman,Telling lies: Clues to deceit in the marketplace, politics, and marriage (revised edition)

    P . Ekman,Telling lies: Clues to deceit in the marketplace, politics, and marriage (revised edition). WW Norton & Company, 2009

  3. [3]

    Cross-domain facial expression recognition: A unified evaluation benchmark and adversarial graph learning,

    T. Chen, T. Pu, H. Wu, Y. Xie, L. Liu, and L. Lin, “Cross-domain facial expression recognition: A unified evaluation benchmark and adversarial graph learning,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 12, pp. 9887–9903, 2022

  4. [4]

    Fa- cial action unit detection by adaptively constraining self-attention and causally deconfounding sample,

    Z. Shao, H. Zhu, Y. Zhou, X. Xiang, B. Liu, R. Yao, and L. Ma, “Fa- cial action unit detection by adaptively constraining self-attention and causally deconfounding sample,”International Journal of Com- puter Vision, vol. 133, no. 4, pp. 1711–1726, 2025

  5. [5]

    How fast are the leaked facial expressions: The duration of micro-expressions,

    W.-J. Yan, Q. Wu, J. Liang, Y.-H. Chen, and X. Fu, “How fast are the leaked facial expressions: The duration of micro-expressions,” Journal of Nonverbal Behavior, vol. 37, no. 4, pp. 217–230, 2013

  6. [6]

    His- tograms of oriented optical flow and binet-cauchy kernels on non- linear dynamical systems for the recognition of human actions,

    R. Chaudhry, A. Ravichandran, G. Hager, and R. Vidal, “His- tograms of oriented optical flow and binet-cauchy kernels on non- linear dynamical systems for the recognition of human actions,” in IEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2009, pp. 1932–1939

  7. [7]

    Dynamic texture recognition using local binary patterns with an application to facial expressions,

    G. Zhao and M. Pietikainen, “Dynamic texture recognition using local binary patterns with an application to facial expressions,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 29, no. 6, pp. 915–928, 2007

  8. [8]

    Micro-facial movement detection using individualised baselines and histogram-based de- scriptors,

    A. K. Davison, M. H. Yap, and C. Lansley, “Micro-facial movement detection using individualised baselines and histogram-based de- scriptors,” inIEEE International Conference on Systems, Man, and Cybernetics. IEEE, 2015, pp. 1864–1869

Show all 69 references
  1. [9]

    Towards reading hidden emotions: A compara- tive study of spontaneous micro-expression spotting and recog- nition methods,

    X. Li, X. Hong, A. Moilanen, X. Huang, T. Pfister, G. Zhao, and M. Pietik ¨ainen, “Towards reading hidden emotions: A compara- tive study of spontaneous micro-expression spotting and recog- nition methods,”IEEE Transactions on Affective Computing, vol. 9, no. 4, pp. 563–577, 2018

  2. [10]

    Less is more: Micro-expression recognition from video using apex frame,

    S.-T. Liong, J. See, K. Wong, and R. C.-W. Phan, “Less is more: Micro-expression recognition from video using apex frame,”Sig- nal Processing: Image Communication, vol. 62, pp. 82–92, 2018

  3. [11]

    Enriched long- term recurrent convolutional network for facial micro-expression recognition,

    H.-Q. Khor, J. See, R. C. W. Phan, and W. Lin, “Enriched long- term recurrent convolutional network for facial micro-expression recognition,” inIEEE International Conference on Automatic Face & Gesture Recognition. IEEE, 2018, pp. 667–674

  4. [12]

    Sponta- neous facial micro-expression recognition using 3d spatiotemporal convolutional neural networks,

    S. P . T. Reddy, S. T. Karri, S. R. Dubey, and S. Mukherjee, “Sponta- neous facial micro-expression recognition using 3d spatiotemporal convolutional neural networks,” inInternational Joint Conference on Neural Networks. IEEE, 2019, pp. 1–8

  5. [13]

    Learning from macro- expression: a micro-expression recognition framework,

    B. Xia, W. Wang, S. Wang, and E. Chen, “Learning from macro- expression: a micro-expression recognition framework,” inACM International Conference on Multimedia, 2020, pp. 2936–2944

  6. [14]

    Learning effective binary descriptors for micro-expression recognition transferred by macro-information,

    X. Ben, X. Jia, R. Yan, X. Zhang, and W. Meng, “Learning effective binary descriptors for micro-expression recognition transferred by macro-information,”Pattern Recognition Letters, vol. 107, pp. 50–58, 2018

  7. [15]

    Efficient spatio- temporal local binary patterns for spontaneous facial micro- expression recognition,

    Y. Wang, J. See, R. C.-W. Phan, and Y.-H. Oh, “Efficient spatio- temporal local binary patterns for spontaneous facial micro- expression recognition,”PLOS ONE, vol. 10, no. 5, p. e0124674, 2015

  8. [16]

    Fuzzy histogram of optical flow orientations for micro-expression recognition,

    S. Happy and A. Routray, “Fuzzy histogram of optical flow orientations for micro-expression recognition,”IEEE Transactions on Affective Computing, vol. 10, no. 3, pp. 394–406, 2019

  9. [17]

    A novel graph-tcn with a graph structured representation for micro-expression recognition,

    L. Lei, J. Li, T. Chen, and S. Li, “A novel graph-tcn with a graph structured representation for micro-expression recognition,” in ACM International Conference on Multimedia, 2020, pp. 2237–2245

  10. [18]

    A novel micro-expression recognition approach using attention- based magnification-adaptive networks,

    M. Wei, W. Zheng, Y. Zong, X. Jiang, C. Lu, and J. Liu, “A novel micro-expression recognition approach using attention- based magnification-adaptive networks,” inIEEE International Conference on Acoustics, Speech and Signal Processing. IEEE, 2022, pp. 2420–2424

  11. [19]

    Multi- scale promoted self-adjusting correlation learning for facial action unit detection,

    X. Liu, K. Yuan, X. Niu, J. Shi, Z. Yu, H. Yue, and J. Yang, “Multi- scale promoted self-adjusting correlation learning for facial action unit detection,”IEEE Transactions on Affective Computing, 2024

  12. [20]

    Auformer: Vision transformers are parameter-efficient facial action unit de- tectors,

    K. Yuan, Z. Yu, X. Liu, W. Xie, H. Yue, and J. Yang, “Auformer: Vision transformers are parameter-efficient facial action unit de- tectors,” inEuropean Conference on Computer Vision. Springer, 2024

  13. [21]

    Au-assisted graph attention convolutional network for micro-expression recognition,

    H.-X. Xie, L. Lo, H.-H. Shuai, and W.-H. Cheng, “Au-assisted graph attention convolutional network for micro-expression recognition,” inACM International Conference on Multimedia. ACM, 2020, pp. 2871–2880

  14. [22]

    Learnet: Dynamic imaging network for micro expression recognition,

    M. Verma, S. K. Vipparthi, G. Singh, and S. Murala, “Learnet: Dynamic imaging network for micro expression recognition,” IEEE Transactions on Image Processing, vol. 29, pp. 1618–1627, 2020

  15. [23]

    Geme: Dual-stream multi-task gender-based micro-expression recogni- tion,

    X. Nie, M. A. Takalkar, M. Duan, H. Zhang, and M. Xu, “Geme: Dual-stream multi-task gender-based micro-expression recogni- tion,”Neurocomputing, vol. 427, pp. 13–28, 2021

  16. [24]

    Dual-inception network for cross- database micro-expression recognition,

    L. Zhou, Q. Mao, and L. Xue, “Dual-inception network for cross- database micro-expression recognition,” inIEEE International Con- ference on Automatic Face & Gesture Recognition. IEEE, 2019, pp. 1–5

  17. [25]

    Identity- invariant representation and transformer-style relation for micro- expression recognition,

    Z. Shao, F. Li, Y. Zhou, H. Chen, H. Zhu, and R. Yao, “Identity- invariant representation and transformer-style relation for micro- expression recognition,”Applied Intelligence, vol. 53, pp. 19 860– 19 871, 2023

  18. [26]

    Multi-task micro- expression recognition combining deep and handcrafted features,

    C. Hu, D. Jiang, H. Zou, X. Zuo, and Y. Shu, “Multi-task micro- expression recognition combining deep and handcrafted features,” inInternational Conference on Pattern Recognition. IEEE, 2018, pp. 946–951

  19. [27]

    Handbook of digital signal processing: Engineering applications,

    D. F. Elliott, “Handbook of digital signal processing: Engineering applications,” 1987

  20. [28]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin, “Attention is all you need,” inAdvances in Neural Information Processing Systems. Curran Associates, Inc., 2017, pp. 5998–6008

  21. [29]

    Spectral networks and deep locally connected networks on graphs,

    J. Bruna, W. Zaremba, A. Szlam, and Y. LeCun, “Spectral networks and deep locally connected networks on graphs,” inInternational Conference on Learning Representations, 2014

  22. [30]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” inIEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2016, pp. 770–778

  23. [31]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, J. Uszkoreit, and N. Houlsby, “An image is worth 16x16 words: Transformers for image recognition at scale,” inInternational Con- ference on Learnin...

  24. [32]

    Interpretable convolutional neural networks,

    Q. Zhang, Y. N. Wu, and S.-C. Zhu, “Interpretable convolutional neural networks,” inIEEE Conference on Computer Vision and Pat- tern Recognition. IEEE, 2018, pp. 8827–8836

  25. [33]

    Dynamic graph cnn for learning on point clouds,

    Y. Wang, Y. Sun, Z. Liu, S. E. Sarma, M. M. Bronstein, and J. M. Solomon, “Dynamic graph cnn for learning on point clouds,”Acm Transactions On Graphics, vol. 38, no. 5, pp. 1–12, 2019

  26. [34]

    Nearest neighbor pattern classification,

    T. Cover and P . Hart, “Nearest neighbor pattern classification,” IEEE Transactions on Information Theory, vol. 13, no. 1, pp. 21–27, 1967

  27. [35]

    Rectified linear units improve restricted boltzmann machines,

    V . Nair and G. E. Hinton, “Rectified linear units improve restricted boltzmann machines,” inInternational Conference on Machine Learn- ing, 2010, pp. 807–814

  28. [36]

    Flownet: Learning optical flow with convolutional networks,

    A. Dosovitskiy, P . Fischer, E. Ilg, P . Hausser, C. Hazirbas, V . Golkov, P . Van Der Smagt, D. Cremers, and T. Brox, “Flownet: Learning optical flow with convolutional networks,” inIEEE International Conference on Computer Vision, 2015, pp. 2758–2766. IEEE TRANSACTIONS ON PA...

  29. [37]

    Deep multi-center learning for face alignment,

    Z. Shao, H. Zhu, X. Tan, Y. Hao, and L. Ma, “Deep multi-center learning for face alignment,”Neurocomputing, vol. 396, pp. 477– 486, 2020

  30. [38]

    J ˆaa-net: Joint facial action unit detection and face alignment via adaptive attention,

    Z. Shao, Z. Liu, J. Cai, and L. Ma, “J ˆaa-net: Joint facial action unit detection and face alignment via adaptive attention,”International Journal of Computer Vision, vol. 129, no. 2, pp. 321–340, 2021

  31. [39]

    Casme ii: An improved spontaneous micro-expression database and the baseline evaluation,

    W.-J. Yan, X. Li, S.-J. Wang, G. Zhao, Y.-J. Liu, Y.-H. Chen, and X. Fu, “Casme ii: An improved spontaneous micro-expression database and the baseline evaluation,”PLOS ONE, vol. 9, no. 1, p. e86041, 2014

  32. [40]

    Samm: A spontaneous micro-facial movement dataset,

    A. K. Davison, C. Lansley, N. Costen, K. Tan, and M. H. Yap, “Samm: A spontaneous micro-facial movement dataset,”IEEE Transactions on Affective Computing, vol. 9, no. 1, pp. 116–129, 2016

  33. [41]

    A spontaneous micro-expression database: Inducement, collection and baseline,

    X. Li, T. Pfister, X. Huang, G. Zhao, and M. Pietik ¨ainen, “A spontaneous micro-expression database: Inducement, collection and baseline,” inIEEE International Conference and Workshops on Automatic Face and Gesture Recognition. IEEE, 2013, pp. 1–6

  34. [42]

    Dlib-ml: A machine learning toolkit,

    D. E. King, “Dlib-ml: A machine learning toolkit,”Journal of Machine Learning Research, vol. 10, pp. 1755–1758, 2009

  35. [43]

    One millisecond face alignment with an ensemble of regression trees,

    V . Kazemi and J. Sullivan, “One millisecond face alignment with an ensemble of regression trees,” inIEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2014, pp. 1867–1874

  36. [44]

    A duality based approach for realtime tv-l 1 optical flow,

    C. Zach, T. Pock, and H. Bischof, “A duality based approach for realtime tv-l 1 optical flow,” inJoint Pattern Recognition Symposium. Springer, 2007, pp. 214–223

  37. [45]

    Off- apexnet on micro-expression recognition system,

    Y. S. Gan, S.-T. Liong, W.-C. Yau, Y.-C. Huang, and L.-K. Tan, “Off- apexnet on micro-expression recognition system,”Signal Process- ing: Image Communication, vol. 74, pp. 129–139, 2019

  38. [46]

    Micro-expression classification based on landmark relations with graph attention convolutional network,

    A. J. R. Kumar and B. Bhanu, “Micro-expression classification based on landmark relations with graph attention convolutional network,” inIEEE Conference on Computer Vision and Pattern Recog- nition Workshops. IEEE, 2021, pp. 1511–1520

  39. [47]

    Micro-expression recognition with supervised contrastive learning,

    R. Zhi, J. Hu, and F. Wan, “Micro-expression recognition with supervised contrastive learning,”Pattern Recognition Letters, vol. 163, pp. 25–31, 2022

  40. [48]

    Sparsity in dynamics of spontaneous subtle emotions: analysis and application,

    A. C. Le Ngo, J. See, and R. C.-W. Phan, “Sparsity in dynamics of spontaneous subtle emotions: analysis and application,”IEEE Transactions on Affective Computing, vol. 8, no. 3, pp. 396–411, 2017

  41. [49]

    Dual- stream shallow networks for facial micro-expression recognition,

    H.-Q. Khor, J. See, S.-T. Liong, R. C. Phan, and W. Lin, “Dual- stream shallow networks for facial micro-expression recognition,” inIEEE International Conference on Image Processing. IEEE, 2019, pp. 36–40

  42. [50]

    Joint local and global information learning with single apex frame detection for micro-expression recognition,

    Y. Li, X. Huang, and G. Zhao, “Joint local and global information learning with single apex frame detection for micro-expression recognition,”IEEE Transactions on Image Processing, vol. 30, pp. 249–263, 2021

  43. [51]

    Micro-expression recognition based on facial graph representation learning and facial action unit fusion,

    L. Lei, T. Chen, S. Li, and J. Li, “Micro-expression recognition based on facial graph representation learning and facial action unit fusion,” inIEEE Conference on Computer Vision and Pattern Recognition Workshops. IEEE, 2021, pp. 1571–1580

  44. [52]

    A two-stage 3d cnn based learning method for spontaneous micro- expression recognition,

    S. Zhao, H. Tao, Y. Zhang, T. Xu, K. Zhang, Z. Hao, and E. Chen, “A two-stage 3d cnn based learning method for spontaneous micro- expression recognition,”Neurocomputing, vol. 448, pp. 276–289, 2021

  45. [53]

    Short and long range relation based spatio-temporal transformer for micro- expression recognition,

    L. Zhang, X. Hong, O. Arandjelovi ´c, and G. Zhao, “Short and long range relation based spatio-temporal transformer for micro- expression recognition,”IEEE Transactions on Affective Computing, vol. 13, no. 4, pp. 1973–1985, 2022

  46. [54]

    Dynamic micro-expression recognition using knowledge distillation,

    B. Sun, S. Cao, D. Li, J. He, and L. Yu, “Dynamic micro-expression recognition using knowledge distillation,”IEEE Transactions on Affective Computing, vol. 13, no. 2, pp. 1037–1043, 2022

  47. [55]

    Megc 2019– the second facial micro-expressions grand challenge,

    J. See, M. H. Yap, J. Li, X. Hong, and S.-J. Wang, “Megc 2019– the second facial micro-expressions grand challenge,” inIEEE International Conference on Automatic Face & Gesture Recognition. IEEE, 2019, pp. 1–5

  48. [56]

    Shallow triple stream three-dimensional cnn (ststnet) for micro- expression recognition,

    S.-T. Liong, Y. S. Gan, J. See, H.-Q. Khor, and Y.-C. Huang, “Shallow triple stream three-dimensional cnn (ststnet) for micro- expression recognition,” inIEEE International Conference on Auto- matic Face & Gesture Recognition. IEEE, 2019, pp. 1–5

  49. [57]

    Pytorch: An imperative style, high-performance deep learning library,

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, A. Desmaison, A. Kopf, E. Yang, Z. DeVito, M. Raison, A. Tejani, S. Chilamkurthy, B. Steiner, L. Fang, J. Bai, and S. Chintala, “Pytorch: An imperative style, high-pe...

  50. [58]

    Adam: A method for stochastic optimiza- tion,

    D. P . Kingma and J. Ba, “Adam: A method for stochastic optimiza- tion,” inInternational Conference on Learning Representations, 2015

  51. [59]

    Expression, affect, action unit recogni- tion: Aff-wild2, multi-task learning and arcface,

    D. Kollias and S. Zafeiriou, “Expression, affect, action unit recogni- tion: Aff-wild2, multi-task learning and arcface,” inBritish Machine Vision Conference. BMVA Press, 2019, p. 297

  52. [60]

    Analysing affective behavior in the second abaw2 compe- tition,

    ——, “Analysing affective behavior in the second abaw2 compe- tition,” inIEEE International Conference on Computer Vision Work- shops. IEEE, 2021, pp. 3652–3660

  53. [61]

    Capsulenet for micro- expression recognition,

    N. Van Quang, J. Chun, and T. Tokuyama, “Capsulenet for micro- expression recognition,” inIEEE International Conference on Auto- matic Face & Gesture Recognition. IEEE, 2019, pp. 1–7

  54. [62]

    A neural micro- expression recognizer,

    Y. Liu, H. Du, L. Zheng, and T. Gedeon, “A neural micro- expression recognizer,” inIEEE International Conference on Auto- matic Face & Gesture Recognition. IEEE, 2019, pp. 1–4

  55. [63]

    Feature representation learning with adaptive displacement generation and transformer fusion for micro-expression recognition,

    Z. Zhai, J. Zhao, C. Long, W. Xu, S. He, and H. Zhao, “Feature representation learning with adaptive displacement generation and transformer fusion for micro-expression recognition,” inIEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2023, pp. 22 086–22 095

  56. [64]

    Selfme: Self- supervised motion learning for micro-expression recognition,

    X. Fan, X. Chen, M. Jiang, A. R. Shahid, and H. Yan, “Selfme: Self- supervised motion learning for micro-expression recognition,” in IEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2023, pp. 13 834–13 843

  57. [65]

    Mer-gcn: Micro- expression recognition based on relation modeling with graph convolutional networks,

    L. Lo, H.-X. Xie, H.-H. Shuai, and W.-H. Cheng, “Mer-gcn: Micro- expression recognition based on relation modeling with graph convolutional networks,” inIEEE Conference on Multimedia Infor- mation Processing and Retrieval. IEEE, 2020, pp. 79–84

  58. [66]

    Back to basics: Un- supervised learning of optical flow via brightness constancy and motion smoothness,

    J. J. Yu, A. W. Harley, and K. G. Derpanis, “Back to basics: Un- supervised learning of optical flow via brightness constancy and motion smoothness,” inEuropean Conference on Computer Vision Workshops, 2016

  59. [67]

    Raft: Recurrent all-pairs field transforms for optical flow,

    Z. Teed and J. Deng, “Raft: Recurrent all-pairs field transforms for optical flow,” inEuropean Conference on Computer Vision. Springer, 2020, pp. 402–419

  60. [68]

    Learning deep rep- resentation for face alignment with auxiliary attributes,

    Z. Zhang, P . Luo, C. C. Loy, and X. Tang, “Learning deep rep- resentation for face alignment with auxiliary attributes,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 38, no. 5, pp. 918–930, 2016

  61. [69]

    Deep high-resolution representa- tion learning for visual recognition,

    J. Wang, K. Sun, T. Cheng, B. Jiang, C. Deng, Y. Zhao, D. Liu, Y. Mu, M. Tan, X. Wanget al., “Deep high-resolution representa- tion learning for visual recognition,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 43, no. 10, pp. 3349–3364, 2021. Zhiwen Sha...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.